Etude Cass sur la rémunération des dirigeants des "entreprises du vice"

Michal

Informations

• Publié par: L_Calixte
• Publié le: 25 février 2015
• Nombre de lectures: 857
• Langue: English

Extrait

Social Stigma and Executive Remuneration:

The Compensation Premium in “Sin” Industries

* Jiri Novak and Pawel Bilinski

1 December 2014

Abstract

We document that executives in alcohol, gambling, and tobacco industries (sin industries) earn a statistically and economically significant compensation premium. Compensation premium in sin industries cannot be attributed to (1) higher complexity of sin industries, which requires more skilled managers, (2) higher compensation risk, (3) higher executive entrenchment, or (4) higher likelihood of dismissal due to poor performance. Rather, the premium compensates for social stigma related to employment in sin industries. Consistent with this prediction, we document that (1) executive compensation in sin industries is higher in periods and in states with higher social aversion to sin and (2) sin firms executives hold less outside board seats, which indicates their lower social prestige. Together, our evidence suggests that executives at sin firms demand a compensation premium for bearing negative personal and professional costs of working in industries perceived negatively in light of prevailing social norms.

Keywords: Executive compensation, sin firms, social norms

JEL classification: G11, D71

* We would like to thankMichal Bauer, Aleš Čornanič, Michael Erkens, YingGan, Mattias Hamberg, Bjorn Jorgensen, andLucie Mišičováfor helpful comments and suggestions. We would like to thank Michael Janský for dedicated research assistantship in the earlier phase of the project. Jiri Novak (email:[e-mail] ) is from Institute of Economic Studies, Faculty of Social Sciences, Charles University in Prague. Pawel Bilinski (email: [e-mail])from Cass Business School, City University London. Jiri Novak gratefully is acknowledges financial support from the Czech Science Foundation, grant number P403/12/2166.

Electronic copy available at: http://ssrn.com/abstract=2470393

I Social Stigma and Executive Remuneration:

The Compensation Premium in “Sin” Industries

…

Abstract

We document that executives in alcohol, gambling, and tobacco industries (sin industries) earn a statistically and economically significant compensation premium. Compensation premium in sin industries cannot be attributed to (1) higher complexity of sin industries, which requires more skilled managers, (2) higher compensation risk, (3) higher executive entrenchment, or (4) higher likelihood of dismissal due to poor performance. Rather, the premium compensates for social stigma related to employment in sin industries. Consistent with this prediction, we document that (1) executive compensation in sin industries is higher in periods and in states with higher social aversion to sin and (2) sin firms executives hold less outside board seats, which indicates their lower social prestige. Together, our evidence suggests that executives at sin firms demand a compensation premium for bearing negative personal and professional costs of working in industries perceived negatively in light of prevailing social norms.

Keywords: Executive compensation, sin firms, social norms

JEL classification: G11, D71

Electronic copy available at: http://ssrn.com/abstract=2470393

2

Introduction

1 This study examines how social norms and perceptions affect executive remuneration contracts. Social economics literature has long recognized that social norms shape economic behavior, e.g. through the effect social norms have on acceptance of racial discrimination at workplace (Altonji & Blank 1999; Levitt 2004). However, there is little evidence on how negative social perception affects executive compensation. We address this question by examining executive compensation in “sin” industries: alcohol, gambling, and tobacco. These industries have long been perceived to 2 violate social norms as their products are harmful to consumers’ physical and mental health. We propose that negative public perception of sin industries can adversely affect social recognition of their executives, which may affect their utility both directly by impairing their social status and indirectly by limiting their opportunities to earn from lucrative outside appointments. If an executive is shunned by the society he or she will receive less invitations to boards of other firms, which are earmarks of social status. At the same time a sin firm executive forgoes income from board memberships that may be significant. Thus, we expect managers in sin industries to demand a compensation premium for the negative consequences sin industries stigma have on their personal and professional life. This study is first to examine how negative social perception of the nature of economic activities in a firm affect executive compensation.

Our empirical tests identify a significant premium in executive compensation of sin firm: an executive in these industries can expect to earn $331,300 more in annual inflation-adjusted income, compared to non-sin industry executives. The premium is paid to both CEOs and other executives, and is present in all components of total remuneration (salary, bonus and equity-based compensation). The premium is largest in the tobacco industry, which is arguably the most stigmatized (Beneish et al. 2008; Gerstein et al. 2004; Anielski & Braaten 2008), followed by the gambling and the alcohol industry. Finally, we document that the premium is not sensitive to alternative definitions of sin industries, subsample splits, and a number of other robustness tests.

1 We follow Akerlof (1980) and Hong and Kacperczyk (2009) and define social norms as behavior where utility is dependent on beliefs or actions of other members of the community. 2 Early codification of social norms on consumption of alcohol and gambling are found in many Christian, Hindu and Islam texts (Fam et al. 2004). Negative public attitude towards tobacco relate to medical evidence in 1960s on the link between cigarette smoking and cancer. Social stigma stems from addictive and pathological effects sin industry products have on consumers, their families and communities (Gerstein et al. 2004; Anielski & Braaten 2008; Leventis et al. 2013; Galvin et al. 2004; Grinols 2004; Hudson 2008). 3

We argue that remuneration premium compensates for adverse effects that social stigma of working in the sin industry has on manager’s personal and professional life. To support this proposition we provide two corroborating pieces of evidence. First, we show that the premium increases following periods of heightened negative social attitude to sin industries, which we proxy by (1) state-level smoking prevention spending per capita and (2) large legal settlements in the tobacco industry. This implies that executives demand higher compensations in places and in times of higher negative public pressures on sin industries. Second, we examine the number of personal connections available to sin executives through seats on boards of directors of other firms. We find that executives from sin firms sit on fewer outside boards and on smaller boards. As board seats are an earmark of social status (Kaplan and Reishus, 1990; Maug et al., 2012), our evidence suggests that 3 stigmatized executives of sin firms receive them less often. Together, our evidence suggests that sin firms pay a compensation premium to their executives to compensate for social stigma working in sin industries entails.

We perform a number of additional tests to ensure our conclusions are not driven by confounding effects. First, we show that the compensation premium cannot be attributed to higher complexity and operating risk in sin industries, which requires appointments of more skilled managers who demand higher compensation. Specifically, following Demerjian et al. (2012), we calculate the firm-level measure of the average ability of the executive team and show that it does not explain the compensation premium. Second, executive compensation at sin firms does not exhibit signs of greater income risk, which could explain the premium. Specifically, there is no evidence that (1) managers in sin industries are more likely to be let go because of poor performance compared to other industries, and (2) the pay-performance sensitivity of executives at sin firms is not different than that of other firms. Third, the compensation premium in sin industries is not due to higher entrenchment of executives, thus their ability to extract a premium when negotiating their employment contracts. Together, our results provide strong support for the explanation that social norms affect executive compensation at sin firms.

This study offers an important contribution to the literature on the determinants of executive compensation as well as to the literature on the impact of social norms on economic activities.

3 The prediction that certain firms do not wish to be associated with executives from sin firms is consistent with the evidence in Hong and Kacperczyk (2009) that a significant number of institutional investors avoid investing in sin industries. 4

Standard compensation theories ignore the effect social context has on managerial compensation contracts. We show that higher executive compensation in sin firms compensates for the social stigma executives at these firms bear. Our evidence is important as Graham et al. (2012) highlight that a large proportion of the cross-sectional variation in executive compensation remains still unexplained. Our evidence complements the results in Maug et al. (2012), who document that CEOs of companies ranked as prestigious by specialized industry press earn lower compensation than executives at non-prestigious firms. The authors argue their evidence reflects that social status garnered by working for a prestigious firm has value to CEOs that compensates for their lower financial remuneration package. Our emphasis on the compensation premium in sin industries highlights how negative social perception of the nature of economic activities in these industries affect executive compensation. Further, our study adds new evidence to the emerging literature that examines how social norms affect firm economic performance (Chong et al., 2006; McGuire et al., 2012; Hong and Kacperczyk, 2009; Leventis et al., 2013). We highlight a new link between the social context a firm operates in and economic outcomes—the labor force compensation premium in firms engaging in activities deemed socially undesirable. The remainder of this paper is organized as follows: Section 2 reviews prior research and develops hypotheses, Section 3 presents the research design, Section 4 describes data, Section 5 discusses results, and Section 6 concludes.

1.Literature review and hypotheses

Only a handful of studies to date examined the impact of social norms on the economic activity in industries that have a negative social perception. Hong and Kacperczyk (2009) investigate the value and ownership structure of sin firms. They find that firms in the tobacco, gambling, and alcohol industries are valued consistently lower than their returns would imply, and that these firms have smaller ownership by norm-constrained institutions such as pension funds, and less analyst following. Chong et al. (2006) and Salaber (2009) also report that institutional investors underinvest in sin firm because of the social stigma such investments entail. The evidence of underpricing and lower institution ownership suggests that social norms lead to significant costs for sin firms.

5

Leventis et al. (2013) find that sin firms pay higher fees for external audit. They propose three alternative explanations for their results: (1) auditors work more diligently on sin firm contracts because the costs of failure in the form of reputation loss for the auditor is greater, (2) auditors do not work harder but extract (an insurance-like) premium to insulate themselves from reputation costs, and (3) sin firms themselves demand more rigorous audit in order to assure the public of the quality of their reporting and thus improve their public perception. Beneish et al. (2008) document that tobacco firms engage in acquisitions to protect against expropriation and litigation by public authorities and private claimants. Acquisitions help (1) divest firm excess cash, which is easier to claim then physical assets in the event of an expropriation attempt, and (2) they expand the firms’ political connections and influence, which can reduce the likelihood of expropriation. Together, these studies provide early evidence that social norms can lead to costly structural and operational changes at sin firms, particularly among tobacco firms. Kim and Venkatachalam (2011) report that sin firms have higher quality financial reporting, which they attribute to higher litigation and regulatory risk these firms face.

Maug et al. (2012) examine compensation of CEOs at firms identified as “prestigious” by their placement in several rankings in business and specialized press. They find that CEOs of prestigious firms are willing to accept lower compensation, which they attribute to non-monetary benefits CEOs gain from working for prestigious firms and improved career opportunities. Maug et al. (2012) do not examine if executives demand a premium for bearing negative social costs of working at firms deemed to break social norms.Ex-ante, it is unclear if compensation premium at sin firms should exist in competitive labor markets. First, country-wide and international job markets allow sin industries to attract executives who, at a margin, do not demand a compensation premium for social stigma. Second, compensation premium that does not reward better skill or performance means a wealth transfer from shareholders to executives. In competitive financial markets shareholders could (1) penalize board of directors that allow wealth transfer to executives or (2) choose alternative investments with same expected return but absent wealth transfers. Thus, in contrast to lower executive compensation that benefits shareholders at prestigious firms, it is unclear if social norms are strong enough to induce a compensation premium in sin industries.

Prior research suggests that social norms have significant economic impact. We expand this literature by examining how social norms affect executive compensation at sin firms. We expect

6

that firms that violate social norms suffer from a loss of social status, which in return adversely affects executives’ personal utility (through a social stigma these executives face) and less developed personal networks. Higher salary at sin firms compensates executives for these negative effects of social stigma. This prediction leads to our first empirical hypothesis:

Hypothesis 1: There is a premium in executive compensation at sin firms.

Previous evidence suggests that tobacco firms face a higher level of public animosity (Beneish et al., 2008; Hong and Kacperczyk, 2009), which can increase the average premium tobacco firms have to pay to their managers. Research and anecdotal evidence shows increasing pressure of various social, health and governmental bodies to limit the production and distribution of cigarettes 4 in the US . This leads to an extension of our main research hypothesis:

Hypothesis 1a: The premium in executive compensation is larger in tobacco industry compared to other sin industries.

We argue that premium in executive compensation exists because it compensates for the negative effect stigma of sin firms has on personal welfare and social standing. Specifically, sin executives may be less desirable as outside directors. This is because outside directorships is considered a mark of status among executives (Kaplan and Reishus, 1990) and certain firms may prefer to avoid being associated with directors from sin industries. Further, outside directorships brings additional financial compensation, which, can be non-trivial if an individual holds multiple directorships (Yermack, 2004). The negative social stigma related to employment at sin firms can make their executives less attractive as outside directors in other firms, particularly for larger boards. This in return can translate into less personal prestige and lower financial gain. In a similar way, certain firms may be less willing to hire sin executives into their management team to avoid negative social impact such an appointment can have. These predictions lead to two hypotheses:

Hypothesis 2a: Sin firm executives sit less frequently as outside directors on boards of other firms.

Hypothesis 2a: Sin firm executives sit on smaller outside boards.

4 See for example the U.S. Public Health Service Surgeon General report on tobacco use and the health of the American peoplehttp://www.cdc.gov/mmwr/preview/mmwrhtml/rr4916a1.htm 7

2.Research Design

This section first describes our measures of executive compensations, then we present our definitions of sin industries, and finally the regression models we use to test if a compensation premium exists in the sin industry.

3.1 Measures of executive compensation and definitions of the sin industries

We follow prior research (Roulstone, 2003; Gabaix and Landier, 2008; Maug et al., 2012) and use total direct compensation as measured in the ExecuComp database,Comp, as our main compensation measure. We further decompose total executive compensation into salary (Salary), bonus (Bonus), and other direct compensation (ODC) components. This decomposition allows us to examine if the compensation premium is channeled through all components or only specific remuneration elements.

To identify sin industries, we define an indicator variableSIN_1, which takes a value of one if a firm belongs to alcohol, gambling or tobacco industries, and is zero otherwise. Specifically, firms with SIC codes in the range 2100–2199 are allocated to the alcohol industry. Firms with SIC codes in the range 2080–2085 are in the gambling industry, and the tobacco industry includes firms with NAICS codes 7132, 71312, 713210, 71329, 713290, 72112, and 721120.

To sharpen the analysis, we disaggregate the sin group into the three individual industries that constitute theSIN_1sample. Specifically, the indicator variableALCOHOLtakes a value of one for firms in the alcohol industry, and zero otherwise.GAMBLINGis an indicator variable for firms in the gambling industry, andTOBACCOis and indicator variable for firms in the tobacco industry. We expect the compensation premium to be the largest in the tobacco industry, which has been under the strongest public pressure (Hamilton et al. 2002), followed by the gambling and the alcohol industries.

In sensitivity tests, we use two other definitions of the sin industry. Specifically, our second definition of the sin industry (SIN_2) includes all firms classified inSIN_1as well as firms which

8

have at least one segment belonging to the sin industries defined above. This definition of sin industries corresponds to the sample used in Hong and Kacperczyk (2009). Our third definition of the sin industry,SIN_3, enlargesSIN_2by adding firms flagged with 'alcohol concern', 'gambling concern', or 'tobacco concern' in the MSCI ESG STATS database (formerly KLD). We use alternative definitions of sin industries to ensure our main results are not driven by misclassification of sin stocks based on SIC and NAICS codes.

3.2 Control variables

We base our set of control variables on the determinants of executive compensation identified in past research. Previous studies document that executives earn higher compensation when working for larger and faster growing firms (Maug et al. 2012; Gabaix and Landier, 2008; Hartzell and Starks, 2003). As in Maug et al. (2014), we measure firm size by its market capitalization (MV), which is the product of the number of shares outstanding and the closing price at the last trading day of the fiscal year, and by firm total revenue (Sales). We measure firm growth by growth in sales (SALES_GR), which is the ratio of total dollar sales for fiscal yeartover total sales for fiscal yeart-1. We control for firm profitability and return performance because Hartzell and Starks (2003), Engel et al. (2010), and Roulstone (2003) document that executives at better performing stocks earn higher compensation. Return on assets (ROA) measures firm’s profitability and is defined as the ratio of income before extraordinary items to book value of assets. We measure firm return performance by the market-adjusted returns (XRET), which is the difference between the firm’s and the S&P 500 index returns in a fiscal yeart.

Executive remuneration may increase with business risk, which compensates the executive for (1) higher variability in compensation and (2) higher likelihood of bankruptcy and employment termination, which increases the risk of executive’s compensation (Maug et al. 2012, Rousltone 2003). We measure business risk by the firm’s standard deviation of monthly stock returns over fiscal yeart(SD_RET). We distinguish between the CEO and other executives since the former can expect to earn on average higher compensation. Specifically,CEOis an indicator variable that takes a value of one if the executive has been CEO in a fiscal year, and zero otherwise. We include an indicator variable for female executives (Female) as Bertrand and Halock (2001) and Carter et

9

al. (2014) find that females are on average paid less than males. We measure executive tenure at the firm (Tenure) because more tenured executives earn on average higher compensation (Finkelstein and Hambrick 1989). Finally, we include a set of year effects (Year effects) to capture trends in the labor market over time.

We adjust all dollar amounts (compensation, firm size and firm sales) for inflation, with the average value of the US CPI for 1982–1984 being the baseline. All continuous variables are Winsorized at 1% level. As is standard in the literate, compensation and accounting data are for the same fiscal 5 year. Our main model specification is:

�=�+� �+� �+� �+� ��+� ��+ � � � � � � � � � � � �+� _�+� �+� ��+� �_�+ � � � � � � � �� � �� ��+� ∑ �=� � � �  (1) ��+�

The regression standard errors are dual-clustered at the executive and firm level (Petersen, 2009). To facilitate interpretation of the regression results, we report the ‘economic magnitude’ of the estimates as ‘percentage abnormal compensation’ (PAC), which is the mean residual/fitted ratio in an auxiliary regression estimated with the same controls and on the same sample (including any restrictions that may be in effect), but with the sin dummy of interest excluded.PACassumes that the mean of the random noise in residuals is zero and that any nonzero elements in the mean are attributable to the excluded variable (the sin dummy).

Because we expect that social stigma is higher in the tobacco industry compared to alcohol and gambling industries, we run a variation of model (1) where we use indicator variables for the three sin industries instead of the sin dummy. This model’s specification is:

�=�+� ���+� ���+� ���+� �+� �+ � 0 1 2 3 4 5 � +� �+� +� _�+� �+� �+� �_+ 6 7 � 8 9 � 10 � 11 � 12 � 20 ∑ � ��+ (2) �=1 12+� �

We expect that the coefficient on TOBACCO is significant and higher in magnitude compared to the other two sin industries.

5 Our conclusions are unchanged when we use lagged values of independent variables.

10

3.Data

We collect information on executive compensation from ExecuComp. Accounting data is from Compustat, and market data from CRSP. The sample period is 1992–2012 as covered by ExecuComp. As in past studies (e.g. Yermack, 2006), we exclude the financial sector (SIC codes 6000–6999) as the capital structure and compensation rules are different compared to other industries. We drop observations where book equity is negative, as these are typically firms in distress and may be under a non-standard management contracts. We further drop observations where total direct compensation is negative for the year. Our final sample includes 147,284 firm- executive-fiscal year observations, which represents 2,520 firms, and 30,638 executives (not tabulated). Of this, there are 1,929 firm-executive-fiscal year observations in sin industries, 325 in tobacco, 830 in gambling and 774 in alcohol (Table 2).

We present descriptive statistics for the variables in Table 1. Mean and median values are comparable with earlier studies (e.g. Carter et al. 2014). Average total executive compensations in our sample is $1.935m, with the interquartile range of $0.843m. Around 1.31% of the sample observations are executive-firm-years for sin industries, 17.63% of observations are for CEOs and 5.73% are for females. Average executive tenure is close to 5 years, firm size is $1.326b and sales are $1.187b. Average sales growth is close to 14.5% and firm profitability is around 9.4%. Market- adjusted returns calculated over the fiscal year are 7.5% and their volatility equals 2.54%. In unreported result, we find that the correlations between the explanatory variables in Table 1 are on average small, with all correlations comfortably below 0.8, which is the rule-of-thumb level for the

potential multicollinearity problem.

[Insert Table 1 around here]

To shed light on the existence of compensation premium in sin industries, Figure 1 plots total compensations over time for sin and non-sin industries. We observe that executives in sin industries earn a significantly higher total compensation compared to other industries over our sample period. As a first-cut test for weather higher average compensation in sin industries reflects a premium, we also highlight periods following large, highly publicized legal cases against the tobacco industry in

11